gistfile1.txt

// Tell the RecordArchiver about our entries, so we can archive them later.
      RecordArchive ra = new RecordArchive(fs, inUrlsHDFS + "/../archive/incomingUrls");
      ra.addJobPath(new Path(inUrlsHDFS));
      Tap[] sourceFiles = new Tap[ra.getJobFiles().length];
      int i = 0;
      // Add only those files to the job, so we can accept more, while the job is running
      for (FileStatus f : ra.getJobFiles()) {
         LOG.info("Adding input files: " + f.getPath().getName());
         sourceFiles[i++] = new Hfs(new FlatfileSequenceFile(new Fields("key", "value")), f.getPath().toString());
         // sourceFiles[i++] = new Hfs(new TextLine(new Fields("value")), f.getPath().toString());
      }